Assay
A working sales system — and an audit of the evidence the profession is taught from, most of which turns out to be published by someone selling the fix.
The brief
A seller's workspace on one screen: queue on the left, the call in the middle, context on the right. The call tree carries the actual words and branches on what the prospect said rather than which section you clicked — objections fire battlecards with the losing response named, qualification coverage fills in as you talk, and the account files itself into the right queue when you hang up. Underneath it sits the uncomfortable finding the whole thing started from: almost every load-bearing statistic in sales is published by a company selling the remedy that statistic implies, and essentially nobody discloses it.
What shipped
One workspace, five tabs, no navigation — because the thing reps actually hate is being made to go and find things. Queues route by outcome instead of being dragged. Coverage is tracked per channel, since eight calls is not eight touches. A disposition cannot be filed without a dated next step. Every statistic in the research is rendered through one component that refuses to show a claim without six fields, including what the publisher had to gain, and citations resolve from a typed ledger at build time so the prose cannot drift from the audit.
Check the source
Three claims the sales profession repeats as settled. One holds, one holds with a caveat, and one was measured before the smartphone by the company selling the fix. Open them.
- Oldroyd (MIT Sloan) with InsideSales.com, which supplied the data
- Sales engagement software marketed as “built for speed to lead”.
- 6 companies · ~15,000 leads
- 2004 – 2007
- Measured before the smartphone, published by the vendor selling the remedy, and never re-run. I found it presented as current benchmarks on six separate sites. None of them mentioned the dates.
- CEB, later Gartner
- Research advisory at the time. The finding was afterwards built into a training and certification business.
- ~6,000 sales representatives
- around 2011
- Fifteen years old and monetised by its publisher, which are both reasons for care. It survives because modern instrumented call data corroborates the mechanism independently. An old study with a live mechanism is a different thing from a stale number nobody revisited.
- Gong
- Conversation intelligence — the instrument that produced the data.
- 1.8 million opportunities
- ongoing platform data
- The largest sample behind anything I cite, and the one claim that survives every methodology objection I can construct — it is a count of contacts on records, not a judgement about behaviour.
Stack
What it did
Live. Seventeen claims examined, four failed — including two the site had relied on before checking, kept on the page marked as failures rather than quietly deleted. The instrument is complete and usable on a real territory today: nothing is transmitted anywhere, so real prospect names never leave the browser. It was also rebuilt once. The first version was nine routes of argument with tools attached, and it took four screens of scrolling to reach a single control; that shape was wrong and the rewrite is the more useful thing to look at.